[Intermediate] — the other design families worth knowing, in one note, and how to evaluate a chain you’ve never heard of.
Why one note
There are hundreds of chains. Most are EVM forks with different validators and a marketing budget — learning them individually teaches you almost nothing. What’s worth knowing is the handful of genuinely different designs, and a method for assessing the rest quickly.
The genuinely different ones
Cosmos — application-specific chains. Rather than one chain hosting many applications, each application gets its own chain, built with the Cosmos SDK and connected by IBC (Inter-Blockchain Communication).
IBC is the interesting part: a light-client-based messaging protocol where each chain verifies the other’s consensus directly, in-protocol. No multisig, no external committee — which is precisely the model that would have prevented most bridge hacks → bridges. It works because Tendermint gives instant finality, making light-client verification tractable. Trading off: your chain needs its own validator set and its own security budget.
Polkadot — shared security. Parachains lease slots on a central Relay Chain, which validates all of them. You get a custom chain without bootstrapping validators. Elegant, and constrained by a limited number of slots that must be won at auction — a real adoption bottleneck.
Move-based chains: Aptos, Sui. Both from ex-Meta Diem engineers, both using Move — a language where assets are first-class linear types. A Coin resource cannot be copied or accidentally discarded; the type system enforces conservation of value at compile time.
This directly attacks a whole vulnerability class. In Solidity a token balance is just a number in a mapping, and every bug that creates value from nothing is a number being wrong. In Move that’s a type error. Genuinely interesting language design, smaller ecosystem → substructural types.
Sui adds an object-centric model with parallel execution for transactions on independent objects, similar in spirit to Solana’s declared accounts.
Near — sharding, with usable accounts. Actually-deployed sharding, plus human-readable account names and a much better onboarding story than most chains.
Cardano — extended UTXO. UTXO with programmability bolted on, plus a formal-methods-heavy development culture (peer-reviewed papers before implementation). Slow to ship; the eUTXO model has genuinely different properties from both Bitcoin’s and Ethereum’s.
Tron. Worth knowing for one reason: it carries an enormous share of global USDT transfer volume, particularly for remittances in emerging markets, on account of very low fees. Technically unremarkable and highly centralised; practically, a large amount of the world’s actual stablecoin usage happens there → stablecoins.
The EVM chains
BNB Chain, Avalanche C-Chain, Polygon PoS, Fantom, Base, Arbitrum, Optimism — from an application developer’s view these are largely interchangeable: same Solidity, same tooling, same OpenZeppelin.
What actually differs: validator count and who they are; whether it’s a genuine rollup or a sidechain (very different security → rollups); fee levels; and which liquidity and users are already there.
“EVM-compatible” is now table stakes, which is itself the interesting fact: the EVM became the field’s de-facto standard instruction set, so a chain adopting it inherits a decade of tooling instantly. That’s a strong network effect and it’s why so few chains attempt their own VM.
Evaluating an unfamiliar chain — the checklist
Six questions, in order. They resolve most chains in ten minutes:
- How many validators, run by how many independent entities, on what hardware? This is the decentralisation answer, and it’s usually available and usually unflattering
- What’s the consensus, and does it halt or fork under partition? → consensus
- Is it a rollup, a sidechain, or an L1? Marketing blurs this deliberately. A sidechain’s security is its own validators, full stop
- Who can upgrade the chain or its bridge contracts, and is there a timelock?
- What’s the actual, sustained throughput and fee level? Not the benchmark
- Where’s the liquidity and where are the users? A technically superior chain with no ecosystem is a worse place to build
And the one that cuts through most of it: what does this chain do that an existing one doesn’t? For a large majority the honest answer is “the same thing, with a token we issued” — and launching a chain has been a business model more often than a technical necessity.
What’s actually converging
Across all the variety, the field is settling on a few things:
- The EVM as a standard target, with alternative VMs as a deliberate differentiator rather than a default
- Parallel execution via declared state access — Solana, Sui, and Ethereum’s access-list discussions all point the same way
- Proof-of-stake. Proof-of-work outside Bitcoin is essentially finished
- zk proofs for cross-chain verification — the credible answer to the bridge problem
- Rollups and app-chains over monolithic L1s — though Solana keeps that argument genuinely open
Key insight
Chain diversity is mostly business diversity, not technical diversity. The genuinely different designs number maybe half a dozen — Bitcoin’s UTXO, Ethereum’s accounts, Solana’s parallel execution, Move’s linear types, Cosmos’s app-chains with light-client messaging. Everything else is a parameter change. Learn the design families and you can place any new chain in five minutes.
Related
- the trilemma — what they’re all trading
- Solana · Bitcoin
- Rust for web3 — Solana, CosmWasm, ink!
- substructural types — the theory behind Move
Source: [reference] — Aug 2026.
Software engineering
Infrastructure & security
Computer science
Hardware & engineering